This report explores electricity consumption patterns within forecasting, demand and behaviours to tackle if data can show us actionable insights.


  1. Policy Impact: 2020 vs 2019 Intervential anlaysis & SARIMAX

  2. Demand & Economics: Duck curve & the economy - An analysis into electricity consumption

  3. Anomalous Behaviours: An experimental approach to detect change in production activity

Policy Impact: 2020 vs 2019 Intervential anlaysis & SARIMAX

Let’s begin with a look at the day to day difference between 2020 and 2019 across the world adjusted for seasonality. Exploring the COVID-19 responses across the world, it is evident that the difference in electricity consumption drastically drops based on the economic policies placed in each country. For example, Italy who began nation wide lockdowns on March 9, saw the immediate impact in electricity consumption over the next couple days. On March 22nd all factories are closed and all nonessential production is halted in Italy causing a further dip consumption difference. This same story can be told with several other countries. Interestingly, Sweden who chose a mitigation strategy still showed very little difference compared to it’s Scandinavian neighbours Norway and Denmark who too much more preventitive measures.

Intervention analysis aims to accommodate for exigenous forces such as a pandemic with-in it’s modelling in time-series analysis with several methods avaliable for trial, such as R packages ‘changepoint’ and ‘mcp’. However, A forecast will a more useful tool in a practical application of electricity consumption. Hence, I’ve chosen to forecast 2020 and 2019 to compare how predictable electricity consumption is as a micro-study.

The figure above is a time-series of electricity consumption in each country. with an overlay of a SARIMAX model predicting energy consumption in 2019 vs 2020, having been trained on the previous 4 years worth of data. The appendix outlines the accuracy of these models.The objective of this plot and table is to illustrate the predictability of electricity consumption data during a normal year vs a pandemic enduced year. The change is clearly picked up in the drastic changes in accuracy.

Electricity consumption can be attributed to weather, seasonality, business cycles & base load. Weather and seasonality are present moment indicators unless a weather forecast can be added to the model. The temperature causes one to use heating and cooling measures. Business cycles and base load (level of business activity) can be presumptiously reactive to conditions unlike weather and seasonality which are reactive in the moment. Hence, electricity consumption attributed to business cycles and base load can be a forward looking indicator into economic activity.

Demand & Economics: Duck curve & the economy - an analysis into electricity consumption

With this rise in solar energy production anticipation of electricity demand has been on the forefront of utility companies. The duck curve simply is the demand for electricity any given time during the day. The early mornings consist of low energy demands, but as people wake up and busineses begin production the demand rises. Then peaks around sunset before dropping. As you may have guessed the production of energy during the peak hours of the day helps reduce the demand needed. This is evident as the year go on, the dip during mid day drops lower and lower forming a duck like shape. This is more prominent in countries with renewable energy rebates such as Sewden’s tax regulation mechanisms and a subsidy scheme.

There’s two critical issues that rise from this. One, as steeper drops in demand occur more rapid increases in the production are needed during peak hours with no sunlight - this poses a serious burden on power infratructure and diminish efficency. The second issue is that during days of over production in solar energy, grid managers turn off solar panels to prevent overloading the power grid and throw away extra solar energy.

The solution is as battery technology evolves, we’re able to store energy more efficiently and locally reducing waste while moving away from non-renewables.

Anomalous Behaviours: An experimental approach to detect change in production activity

t-SNE (t-distributed Stochastic Neighbor Embedding) is something called nonlinear dimensionality reduction. This is an algorithm which allows us to separate data that cannot be separated by any straight line. Illustrated below are 3 years worth of Electricity Demand Data from Victoria. 2018 in red, 2019 in blue & 2020 in orange. Hovering over the data, we are able to see what factors such as RRP of electricity, week, day & hour impact the cluster formed. Points which appear further away are indicative of outliers. Interestingly, March 2020 appears to be the only occurance where the pattern breaks.

Monthly Electricity Consumption Patterns in Victoria using t-SNE